Framework for generating model-based system control parameters

• US 7,725,199 B2
• Filed: 03/02/2005
• Issued: 05/25/2010
• Est. Priority Date: 03/02/2005
• Status: Expired due to Fees

• Current Assignee: Cummins Incorporated
• Original Assignee: Cummins Incorporated
• Inventors: Brackney, Larry J.
• Primary Examiner(s): Patel; Ramesh B
• Assistant Examiner(s): GAMI, TEJAL
• Application Number:
  US11/070,146
• Publication Number:
  US 20060212140A1
• Time in Patent Office: 1,910 Days
• Field of Search: 700/29, 123/568.21, 606/02
• US Class Current: 700/29
• CPC Class Codes: F02B 29/0493 Controlling the air charge temperature
• CPC Class Codes: F02D 11/105 characterised by the function converting demand to actuation, e.g. a map indicating relations between an accelerator pedal position and throttle valve opening or target engine torque
• CPC Class Codes: F02D 2041/1433 using a model or simulation of the system
• CPC Class Codes: F02D 41/0007 for control of turbo-charged or super-charged engines control of the pumps per se F02B37/12
• CPC Class Codes: F02D 41/005 according to engine operating conditions
• CPC Class Codes: F02D 41/0072 Estimating, calculating or determining the EGR rate, amount or flow sensors in EGR systems F02M26/45
• CPC Class Codes: F02D 41/1401 characterised by the control or regulation method F02D41/1473, F02D41/1477 take precedence
• CPC Class Codes: F02D 41/144 Sensor in intake manifold
• CPC Class Codes: F02D 41/1447 with determination means using an estimation
• CPC Class Codes: F02D 41/1462 with determination means using an estimation
• CPC Class Codes: F02D 41/1467 with determination means using an estimation
• CPC Class Codes: F02M 26/05 High pressure loops, i.e. wherein recirculated exhaust gas is taken out from the exhaust system upstream of the turbine and reintroduced into the intake system downstream of the compressor
• CPC Class Codes: F02M 26/33 controlling the temperature of the recirculated gases
• CPC Class Codes: G05B 13/042 in which a parameter or coefficient is automatically adjusted to optimise the performance
• CPC Class Codes: Y02T 10/12 Improving ICE efficiencies
• CPC Class Codes: Y02T 10/40 Engine management systems

Abstract

A control framework generates control parameters for controlling operation of a physical system, and includes one or more embedded models each producing a model output corresponding to a different operating parameter of the system as a function of one or more system operating conditions and/or a number of solution parameters, objective logic producing a scalar performance metric as a function of the number model outputs and of one or more system performance target values, objective optimization logic determining a number of unconstrained solution parameters in a manner that minimizes the scalar performance metric, and solution constraining logic determining the number of solution parameters from the number of unconstrained solution parameters in a manner that limits an operating range of at least one of the unconstrained solution parameters. The control parameters may correspond to one of the number of unconstrained solution parameters or to the number of solution parameters.

Claims

1. A control framework generating control parameters for controlling operation of a physical system, the control framework comprising:

one or more embedded models each producing a model output corresponding to a different operating parameter of the physical system as a function of either of one or more operating values corresponding to operating conditions of the physical system and one or more of a number of solution parameters, objective logic producing a scalar performance metric as a function of the one or more model outputs, of one or more weight values and of one or more system performance target values, wherein the one or more model outputs defines a vector, Y, the one or more system performance target values defines a vector, Y_T, and the one or more weight values defines a vector, W, and wherein the objective logic is configured to determine a difference vector as a difference between the vectors Y and Y_T, and to determine the scalar performance metric as a vector inner product of the vector W and a function of the difference vector, objective optimization logic producing a number of unconstrained solution parameters in a manner that minimizes the scalar performance metric, and solution constraining logic determining the number of solution parameters from the number of unconstrained solution parameters in a manner that limits an operating range of at least one of the unconstrained control parameters, wherein the control parameters correspond to one of the number of unconstrained control parameters and the number of solution parameters, and wherein the solution constraining logic is further configured to produce at least one of the number of solution parameters as a function of at least one of the one or more system performance target values, and wherein the physical system is an internal combustion engine including an air handling system.

3. The control framework of claim 1 wherein the objective logic is configured to determine the scalar performance metric according to the relationship U=W·

(Y−

Y_T), where U is the scalar performance metric.

4. The control framework of claim 1 wherein the objective logic is configured to determine the scalar performance metric according to the relationship U=W·

(Y−

Y_T)², where U is the scalar performance metric.

5. The control framework of claim 1 wherein the objective logic is configured to determine the scalar performance metric according to the relationship U=W·

|Y−

Y_T|, where U is the scalar performance metric.

6. The control framework of claim 1 wherein the objective logic is configured to determine the scalar performance metric according to the relationship U=W·

|(Y−

Y_T)/Y_T|, where U is the scalar performance metric.

7. The control framework of claim 1 wherein the number of solution parameters define a vector, X, the number of unconstrained solution parameters defines a vector, X′

, and the scalar performance metric is designated U;

and wherein the objective optimization logic is configured to produce X′

as a function of U and X and a specified step size according to a direct search optimization technique.

8. The control framework of claim 7 wherein the objective optimization logic is configured to produce X′

as a function of U and X and a specified step size according to a random walk optimization algorithm.

9. The control framework of claim 7 wherein the objective optimization logic is configured to produce X′

as a function of U and X and a specified step size according to a random walk optimization algorithm with step length adjustment.

10. The control framework of claim 7 wherein the objective optimization logic is configured to produce X′

as a function of U and X and a specified step size according to a random walk optimization algorithm with direction exploitation.

11. The control framework of claim 7 wherein the objective optimization logic is configured to produce X′

as a function of U and X and a specified step size according to a random walk optimization algorithm with direction exploitation and step length adjustment.

12. The control framework of claim 7 wherein the objective optimization logic is configured to produce X′

as a function of U and X and a specified step size according to a variant of a random walk optimization algorithm.

13. The control framework of claim 7 wherein the objective optimization logic is configured to produce X′

as a function of U and X and a specified step size according to a univariate optimization algorithm.

2. A control framework generating control parameters for controlling operation of a physical system, the control framework comprising:

one or more embedded models each producing a model output corresponding to a different operating parameter of the physical system as a function of either of one or more operating values corresponding to operating conditions of the physical system and one or more of a number of solution parameters,

objective logic producing a scalar performance metric as a function of the one or more model outputs, of one or more weight values and of one or more system performance target values, wherein the one or more model outputs defines a vector, Y, the one or more system performance target values defines a vector, Y_T, and the one or more weight values defines a vector, W, and wherein the objective logic is configured to determine a difference vector as a difference between the vectors Y and Y_T, and to determine the scalar performance metric as a vector inner product of the vector W and a function of the difference vector,

objective optimization logic producing a number of unconstrained solution parameters in a manner that minimizes the scalar performance metric, and

solution constraining logic determining the number of solution parameters from the number of unconstrained solution parameters in a manner that limits an operating range of at least one of the unconstrained control parameters,

wherein the control parameters correspond to one of the number of unconstrained control parameters and the number of solution parameters, and wherein the objective optimization logic includes solution selection logic responsive to a plurality of recent iterations of the scalar performance metric and to a corresponding plurality of recent iterations of the number of unconstrained solutions parameters to determine the control parameters as the one of the plurality of recent iterations of the number of unconstrained solution parameters having a corresponding one of the plurality of recent iterations of the scalar performance metric having a minimum magnitude with respect to remaining ones of the plurality of recent iterations of the scalar performance metric, and wherein the physical system is an internal combustion engine including an air handling system.

14. The control framework of claim 2 wherein the control framework is configured to produce a commanded fuel quantity value as one of the control parameters and to produce a commanded start-of-injection value as another one of the control parameters.

15. The control framework of claim 14 further including a fuel system responsive to fueling commands to supply fuel to the engine;

and wherein the control computer includes fueling logic responsive to the commanded fuel quantity value and the commanded start-of-injection value to produce the fueling commands.

16. The control framework of claim 2 wherein the control framework is configured to produce a commanded charge flow value as one of the control parameters and to produce a commanded exhaust gas recirculation (EGR) fraction value as another one of the control parameters.

17. The control framework of claim 16 wherein the air handling system includes an exhaust gas recirculation (EGR) conduit fluidly coupled at one end to an intake manifold of the engine and at an opposite end to an exhaust manifold of the engine, and an EGR valve responsive to an EGR control signal to control the flow of engine exhaust gas through the EGR conduit;

and wherein the control computer includes charge manager logic responsive to the commanded charge flow value and the commanded EGR fraction value to produce the EGR control signal.

18. The control framework of claim 16 wherein the air handling system includes a turbocharger having a variable geometry turbine (VGT) fluidly coupled to an exhaust manifold of the engine, the VGT responsive to a VGT control signal to control the swallowing capacity of the turbine;

and wherein the control computer includes charge manager logic responsive to the commanded charge flow value and the commanded EGR fraction value to produce the VGT control signal.

19. The control framework of claim 16 wherein the air handling system includes an exhaust throttle disposed in-line with an exhaust conduit fluidly coupling an exhaust manifold of the engine to ambient, the exhaust throttle responsive to an exhaust throttle control signal to control engine exhaust gas flow through the exhaust conduit;

and wherein the control computer includes charge manager logic responsive to the commanded charge flow value and the commanded EGR fraction value to produce the VGT control signal.

20. The control framework of claim 2 wherein the number of embedded models includes an engine output torque model producing as a model output an estimate of engine output torque as a function of one or more engine operating parameters.

21. The control framework of claim 2 wherein the number of embedded models includes a peak cylinder pressure model producing as a model output an estimate of peak cylinder pressure as a function of one or more engine operating parameters.

22. The control framework of claim 2 wherein the number of embedded models includes an engine exhaust gas temperature model producing as a model output an estimate of engine exhaust gas temperature as a function of one or more engine operating parameters.

23. The control framework of claim 2 wherein the number of embedded models includes a NOx model producing as a model output an estimate of NOx produced by the engine as a function of one or more engine operating parameters.

24. The control framework of claim 2 wherein the number of embedded models includes a dry particulate matter model producing as a model output an estimate of dry particulate matter produced by the engine as a function of one or more engine operating parameters.

25. The control framework of claim 2 wherein the number of embedded models includes a plurality of fuel limiting models each producing as an output a different fuel flow limit value for limiting engine fueling.

26. The control framework of claim 25 wherein the plurality of fuel limiting models include a peak cylinder pressure (PCP) fuel limit model producing as a model output a PCP-limited fuel flow value as a function of a target PCP limit value included as one of the one or more system performance target values, and as a function of one or more engine operating values.

27. The control framework of claim 25 wherein the plurality of fuel limiting models include an exhaust temperature fuel limit model producing as a model output an exhaust temperature-limited fuel flow value as a function of a target exhaust gas temperature limit value included as one of the one or more system performance target values, and as a function of one or more engine operating parameters.

28. The control framework of claim 25 wherein the plurality of fuel limiting models include a dry particulate matter (DPM) fuel limit model producing as a model output a DPM-limited fuel flow value as a function of a target DPM limit value included as one of the one or more system performance target values, and as a function of one or more engine operating parameters.

29. The control framework of claim 2 wherein the objective logic is configured to determine the scalar performance metric according to the relationship U=W·

|100*(Y_T−

Y)/Y_T|, where U is the scalar performance metric.

30. The control framework of claim 2 wherein the number of solution parameters define a vector, X, the number of unconstrained solution parameters define a vector, X′

, and the scalar performance metric is designated U;

and wherein the objective optimization logic is configured to produce X′

as a function of U and X and a specified step size.

31. The control framework of claim 30 wherein the objective optimization logic is configured to produce X′

as a function of U and X and a specified step size according to a random walk optimization algorithm with direction exploitation and step length adjustment.

32. The control framework of claim 2 wherein the solution constraining logic includes a number of constraint functions each producing specified ones of the number of solution parameters by limiting the specified ones of the corresponding number of unconstrained solution parameters to definable operating ranges.

33. The control framework of claim 32 wherein one of the control parameters is a commanded fuel quantity value;

and wherein the number of constraint functions includes fuel quantity limiting logic limiting the corresponding unconstrained commanded fuel quantity value to a minimum of a maximum torque fueling value, the greater of a minimum torque fueling value and the unconstrained commanded fuel quantity value, a peak cylinder pressure fuel limit value produced by one of the embedded models, an engine exhaust gas temperature fuel limit value produced by another one of the embedded models and a dry particulate matter fuel limit value produced by yet another one of the embedded models.

34. The control framework of claim 33 wherein the fuel quantity limiting logic is configured to determine the maximum and minimum torque fueling values each as a function of a target engine output torque value forming one of the system performance target values and engine speed.

35. The control framework of claim 32 wherein one of the control parameters is a commanded charge flow value and another one of the control parameters is a commanded EGR fraction value;

and wherein the control framework further includes charge management logic responsive to the commanded charge flow value and the commanded EGR fraction value to control one or more actuators associated with the air handling system of the engine;

and wherein the number of constraint functions includes limit accommodation logic limiting the corresponding unconstrained commanded charge flow and commanded EGR fraction values as a function of information fed back to the limit accommodation logic from the charge management logic.

36. The control framework of claim 35 wherein the limit accommodation logic is configured to produce the commanded charge flow value by limiting the corresponding unconstrained commanded charge flow value as a function of charge flow information fed back to the limit accommodation logic from the charge management logic.

37. The control framework of claim 35 wherein the limit accommodation logic is configured to produce the commanded EGR fraction value by limiting the corresponding unconstrained commanded EGR fraction value as a function of EGR fraction information fed back to the limit accommodation logic from the charge management logic.

38. The control framework of claim 33 wherein the fuel quantity limiting logic is further configured to determine an EGR disable value as a function of the unconstrained commanded fuel quantity value, the dry particulate matter fuel limit value and engine speed.

39. The control framework of claim 38 wherein another one of the control parameters is a commanded EGR fraction value;

and wherein the control framework further includes charge management logic responsive to the commanded EGR fraction value to control one or more actuators associated with the air handling system of the engine;

and wherein the control framework further includes limit accommodation logic configured to produce the commanded EGR fraction value by limiting a corresponding unconstrained commanded EGR fraction value as a function of EGR fraction information fed back to the limit accommodation logic from the charge management logic;

and wherein the limit accommodation logic is configured to produce a zero commanded EGR fraction value if the EGR disable value is true, and to otherwise produce the commanded EGR fraction value as long as the commanded EGR fraction value is greater than a minimum EGR fraction value.

40. A method of generating control parameters for controlling operation of a physical system, the method comprising:

maintaining one or more embedded models each producing a model output corresponding to a different operating parameter of the physical system as a function of either of one or more operating values corresponding to operating conditions of the physical system and one or more of a number of solution parameters, producing a scalar performance metric as a function of the one or more model outputs, of one or more weight values and of one or more system performance target values, wherein the one or more model outputs defines a vector, Y, the one or more system performance target values defines a vector, Y_T, and the one or more weight values defines a vector, W, and wherein producing a scalar performance metric comprises determining a difference vector as a difference between the vectors Y and Y_T, and computing the scalar performance metric as a vector inner product of the vector W and a function of the difference vector, generating a number of unconstrained solution parameters in a manner that minimizes the scalar performance metric, determining from the number of unconstrained solution parameters the number of solution parameters in a manner that limits an operating range of at least one of the number of unconstrained solution parameters, and selecting as the control parameters one of the number of solution parameters and the number of unconstrained solution parameters, wherein the step of determining the number of solution parameters includes determining at least one of the number of solution parameters as a function of at least one of the one or more system performance target values, and wherein the physical system is an internal combustion engine including an air handling system.

41. The method of claim 40 wherein the step of producing a scalar performance metric includes computing the scalar performance metric according to the relationship U=W·

(Y−

Y_T), where U is the scalar performance metric.

42. The method of claim 40 wherein the step of producing a scalar performance metric includes computing the scalar performance metric according to the relationship U=W·

(Y−

Y_T)², where U is the scalar performance metric.

43. The method of claim 40 wherein the step of producing a scalar performance metric includes computing the scalar performance metric according to the relationship U=W·

|Y−

Y_T|, where U is the scalar performance metric.

44. The method of claim 40 wherein the step of producing a scalar performance metric includes computing the scalar performance metric according to the relationship U=W·

|(Y−

Y_T)/Y_T|, where U is the scalar performance metric.

45. The method of claim 40 wherein the number of solution parameters define a vector, X, the number of unconstrained solution parameters define a vector, X′

, and the scalar performance metric is designated U;

and wherein the step of generating a number of unconstrained solution parameters includes producing X′

as a function of U and X and a specified step size according to a direct search optimization technique.

46. The method of claim 45 wherein the step of generating a number of unconstrained solution parameters includes producing X′

as a function of U and X and a specified step size according to a random walk optimization algorithm.

47. The method of claim 45 wherein the step of generating a number of unconstrained solution parameters includes producing X′

as a function of U and X and a specified step size according to a random walk optimization algorithm with step length adjustment.

48. The method of claim 45 wherein the step of generating a number of unconstrained solution parameters includes producing X′

as a function of U and X and a specified step size according to a random walk optimization algorithm with direction exploitation.

49. The method of claim 45 wherein the step of generating a number of unconstrained solution parameters includes producing X′

as a function of U and X and a specified step size according to a random walk optimization algorithm with direction exploitation and step length adjustment.

50. The method of claim 45 wherein the step of generating a number of unconstrained solution parameters includes producing X′

as a function of U and X and a specified step size according to a variant of a random walk optimization algorithm.

51. The method of claim 45 wherein the step of generating a number of unconstrained solution parameters includes producing X′

as a function of U and X and a specified step size according to a univariate optimization algorithm.